How to improve failback times for replicas

[ibarizz]

Hello everyone!

I come with a query to address in a scenario that was presented to me with one of my clients. We are testing failovers of our client’s business replicas, the idea is to perform a Failover Now and then carry out the Failback with the commit of the changes made at site B. What we find is that during the failback process, Veeam Backup & Replication needs to calculate and synchronize the differences between the original VM and the replica. This involves scanning the VM image, which can be a time-consuming process depending on the size of the VM and is comparable to the time it would take for a full backup of this VM. In very brief summary, it takes too much time to perform this task.

Reviewing Veeam’s documentation, I found the alternative of Quick Rollback. However, it mentions the following: “Do not use quick rollback if the problem has occurred at the VM hardware level, storage level or due to a power loss.” https://helpcenter.veeam.com/docs/backu ... ml?ver=120

I understand that for our tests where we are simply testing the use of Failover and Failback it will serve us well and will not generate any disruption in the virtual machines. Now, in the event of an event like the one mentioned there, why is it not feasible to use this feature?

Additionally, we want to know if there are ways to improve the failback times that we may not be considering. Thank you very much!

[tyler.jurgens]

The idea of failing back to your production site is meant to be handled on a more controlled manner than failing over in the event of a disaster. Eg: When you have a DR situation, you need to fail over now and get back up and running. However, coming back from the DR site to your primary site can be planned. So you fail back to production while continuing to work on the DR site. When you're ready to make the swap back to the production site, you initiate the final move back which will copy only the changed blocks between when you started the process, and the final click of the button. It shouldn't need to be fast, unless you're DR site is under duress at the same time.